Cold tolerance is a fundamental trait linked to distributions of diverse animals (Andersen et al., 2015, Bishop et al., 2017, Sunday et al., 2019). Variation in cold tolerance among populations and species across space and among individuals through time, from acute responses to acclimation to seasonal variation, can inform our understanding of organismal ecology both now and under changing future climates (Campbell-Staton et al., 2017, Seebacher et al., 2015). Differences in cold tolerance depend on cellular-level mechanisms that are perhaps best understood for insects and provide a compelling opportunity to connect genotype to phenotype within the context of broader ecological implications (Brankatschk et al., 2018, Pimsler et al., 2020, Teets and Denlinger, 2013, Treanore et al., 2020).

Differences in physiological traits underlying cold tolerance variation across populations often have a genetic basis (e.g., Clark and Worland, 2008, Ma et al., 2020, Norry et al., 2007, Rako et al., 2007). Additionally, changes in gene expression have been linked to differences in cold tolerance both in conjunction with genetic differences (e.g., Ma et al., 2020, Parker et al., 2015) and in cases where there is little detectable genetic differentiation among populations (Pimsler et al., 2020, Sørensen et al., 2007). For example, cold-acclimated D. melanogaster exposed to 6 °C for 6 days had enhanced cold tolerance (further depressed CTmin) in conjunction with increased expression of genes previously associated with cold tolerance (i.e., upheld, Tpn25D, Frost, HSP22, SMP-30, etc.; Clowers et al., 2010, MacMillan et al., 2016). During recovery from -10 °C for 2 hours, over a hundred genes related to cytoskeleton structure, organization, and cell shape were differentially expressed in Sarcophaga bullata (Teets et al., 2012) further coupling differential gene expression to cold tolerance and underlying ion homeostasis. Beyond gene expression alone, post-transcriptional changes expressed via proteomes may ultimately influence cold tolerance as elevated gene expression doesn’t always result in changes in proteins (Tomanek, 2011). In lab-reared D. melanogaster selected for stress resistance, 118 proteins differed in abundance between cold shock resistant and control populations (Sørensen et al., 2017).

In concert with or independent of shifts in gene expression, changes in key metabolites that interact with membranes and proteins, thereby influencing cellular function, can also affect cold tolerance (Toxopeus and Sinclair, 2018). For example, sugars, i.e., trehalose (Koštál et al., 2011) and sucrose (Colinet et al., 2012, Kimura, 1982, Olsson et al., 2016), polyols (sorbitol; Colinet et al., 2012), and free amino acids, such as proline (Koštál et al., 2011, MacMillan et al., 2016) and alanine (Olsson et al., 2016), are elevated in Drosophila in response to cold acclimation. Drosophila species that differ in cold tolerance (LTe50: the temperature that results in 50% mortality) had distinct metabolomes after exposure to 0 °C for 4 hours (Olsson et al., 2016). Additionally, cold-hardy D. melanogaster have fewer changes in metabolite concentrations before and during exposure to cold compared to cold-susceptible populations, with hardier populations having lower concentrations of metabolites from the start, except for few that include phosphorylcholine and histidine (Williams et al., 2014).

Although many studies have explored associations between ‘omics profiles and variation in cold tolerance, few have examined how constitutive differences among natural populations expressed in the absence of cold exposure may prime cold-adapted organisms to better tolerate cold (but see López et al., 2002, Martínez-Fernández et al., 2008). Perhaps the best evidence for constitutive priming comes from laboratory studies of D. melanogaster lines artificially selected for cold tolerance. Early work found that cold-selected lines had constitutively higher concentrations of glycogen, triacylglycerols and total protein (Chen and Walker, 1994) as well as of proline (Misener et al., 2001). Other fly lines selected for resistance to cold-shock showed no constitutive differences in gene expression relative to controls (Sørensen et al., 2007); but follow-up proteomic analyses revealed over 100 differentially regulated proteins, including several heat shock proteins, suggesting that constitutive differences in molecular chaperone abundance may facilitate cold resistance (Sørensen et al., 2017). Furthermore, metabolomics revealed constitutively higher maltose and histidine and decreased free amino acids in cold selected lines (Malmendal et al., 2013). And cold hardy lines have been found to have larger, more connected metabolic networks than cold susceptible lines (Williams et al., 2014). Together, these studies suggest that constitutive differences in ‘omics may play an important role in population variation in cold tolerance. However, few studies have measured these differences for wild populations (see López et al., 2002, Martínez-Fernández et al., 2008) while also accounting for potential rearing (e.g., Kristensen et al., 2016) and acclimation (e.g., Colinet et al., 2013) effects, which can be pronounced.

Bumble bees (genus Bombus) are diverse and broadly distributed, with species and populations experiencing pronounced differences in climate. Recent work has found that cold tolerance (CTmin) differs strikingly across populations of the yellow-faced bumble bee (Bombus vosnesenskii; Pimsler et al., 2020), despite little population structure (Jackson et al., 2018), suggesting that cold tolerance is locally adapted. Genomics studies suggest signatures of selection on cold tolerance across the range of the species associated with neural and muscular function, ion transport, signaling, and channel maintenance (Pimsler et al., 2020). Common-garden reared bumble bees reared from queens collected from different populations showed strong gene expression differences in response to cold exposure: northern high-altitude populations upregulated expression of genes linked to membrane fluidity and ion channel function (e.g., flotillin-1, Shal; Pimsler et al., 2020). Similarly, genes related to carbohydrate metabolism were up-regulated in high-elevation species relative to their lower elevation counterparts (Liu et al., 2020). Yet, to what extent differences in cold tolerance across populations may also reflect differences (constitutive and induced) in cellular metabolites is unclear.

Because insect body compartments (head, thorax, and abdomen) differ in functions and associated tissue types, they may differ in cellular composition and metabolites associated with cold tolerance. For example, in the head the brain triggers glycerol production immediately after cold exposure (Yoder et al., 2006). In the thorax, the structure of muscle cellular membranes can vary with temperature, facilitating maintenance of membrane potential (preventing, for example, cold-induced cellular depolarization; Bayley et al., 2020). In the abdomen, ion homeostasis depends on the balance of secretion of ions into the hemolymph by the Malpighian tubules and reabsorption of ions by the hindgut; cold-tolerant species better maintain ion homeostasis in part due to the continued function of these key tissues (Andersen et al., 2017, Overgaard et al., 2021, Yerushalmi et al., 2018). Nevertheless, some tissues are found in multiple compartments. For example, neuromuscular junctions, the function of which strongly affects cold tolerance (reviewed in MacMillan and Sinclair, 2011, Overgaard and MacMillan, 2017), are found throughout the body. Similarly, the fat body, a key regulator of diverse processes as well as primary site of lipid storage, is predominantly found in the abdomen, but also extends to the thorax and head (Chapman, 1998), and hemolymph bathes all tissues (Heinrich, 2004). Similarities and differences in constitutive metabolites across body compartments may therefore suggest some of the processes underlying variation in cold tolerance.

We used a mass spectrometry-based untargeted metabolomics approach to answer the following question: do population differences in thermal tolerance traits reflect, in part, constitutive differences in cellular metabolites? Previous work revealed striking variation in cold tolerance across bumble bee populations (Fig. 1), with common-garden reared workers tolerating ∼6 °C cooler temperatures when reared from queens collected in Oregon relative to those reared from queens collected in California (Pimsler et al., 2020). We captured wild spring B. vosnesenskii queens from these two populations, established colonies in common garden conditions to analyze the metabolomes of workers pulled directly from those colonies. This approach allowed us to detect constitutive differences in metabolomes between populations of bumble bees that differ in cold tolerance without exposure to cold during development or as adults.